Dr. John Trinder
The University of Melbourne, Australia
SLEEP APNEA Authors: John Trinder, Department of Psychology, University of Melbourne Atul Malhotra, Sleep Disorders Program, Brigham and Women’s Hospital, Harvard Medical School.
The disorder, Sleep Apnea, or Sleep Disordered Breathing (SDB), is characterized by the repetitive reduction in airflow during sleep. SDB is recognized as a prevalent disorder with deleterious behavioral, cognitive, cardiovascular and metabolic consequences.
Contents 1. General Features 2. Epidemiology and Risk Factors 3. Pathophysiology 1. Structural Considerations 2. Respiratory Control 4. Consequences 5. Diagnosis 6. Management 7. Bibliography
1. General Features Apnea events are classified according to two features. First, they are described as “apneas” if there is complete cessation of airflow, or as “hypopneas” if there is an important reduction, but not complete cessation, of airflow. Second, they are classified as “obstructive” if there is reduction or cessation of airflow in association with continued respiratory effort, whereas they are called “central” if respiratory events occur in the absence of respiratory effort, and “mixed” if central features are followed by a resumption of respiratory effort before the recommencement of airflow. The extent to which a patient experiences central or obstructive events then defines their particular disorder: Central Sleep Apnea (CSA) or Obstructive Sleep Apnea (OSA). While the different apnea events are conceptually clear, their formal identification is complicated by a number of assessment issues. The first is the extent to which airflow must fall to define a hypopnea. Most definitions accept a fall of either 30% or 50%. However, percent criteria are not precise, as different methods of assessing airflow, for example thermal sensors, nasal pressure transducers, or inductance plethysmography, do not show close agreement with each other. Furthermore, there has not been a consensus as to the preferred method. Thus, the identification of hypopneas is not consistent between laboratories. The second difficulty is that the more accurate methods of identifying respiratory effort are intrusive (e.g. esophageal manometry), making them inappropriate for general clinical use and potentially disruptive to sleep in research studies. As a consequence, the absence of respiratory effort, and thus the presence of central apnea/hypopnea, can be difficult to identify. Indeed, the general recommendation is that hypopneas cannot be reliably distinguished as central or obstructive and thus attempts at this distinction for clinical purposes are discouraged. Finally, there has not been complete consensus as to whether oxygen desaturation and/or electroencephalographic evidence of arousal should be components of the formal definition of a respiratory event. While these are difficult issues, there has been a progressive refinement of definitions over recent years, culminating in the recently published American Association of Sleep Medicine manual (see Bibliography). At present, most experts define apnea as cessation of airflow for at least 10 seconds, which can be obstructive or central depending on evidence for respiratory effort. Hypopneas, on the other hand, are often defined as a >50% reduction in airflow, or a discernable reduction in airflow with associated consequences (either desaturation of 4% or EEG arousal). The severity of the disorder is primarily defined in terms of the frequency of apnea/hypopnea events, although other indices, based on events associated with apneas, may also be quantified. These include the frequency of arousal from sleep, the frequency and magnitude of oxygen desaturation and the quality of sleep. The frequency of respiratory events is quantified as the Apnea-Hyponea Index (AHI), reflecting the average number of respiratory events per hour of sleep. The term AHI is often used interchangeably with Respiratory Disturbance Index (RDI), although technically the latter term also includes more subtle respiratory events that are not of sufficient severity to be classified as hypopneas (e.g. respiratory effort related arousals or upper airways resistance syndrome). However, the AHI can only be considered an approximation of severity and can, under certain circumstances, be misleading. For example, a moderate AHI with long apnea events, and thus greater oxygen desaturation, is likely to be a more deleterious condition than a high AHI with shorter apneas. Generally additional insights have not been gained by separating apneas from hypopneas; therefore, the two types of events are not usually distinguished for clinical purposes. The different indices of SDB, e.g. the arousal and respiratory disturbance indices, do not show particularly strong associations. The extent to which these different events relate to different consequences of SDB will be considered below. Moreover, the AHI is a relatively poor predictor of most sleep apnea complications, leading many to seek new metrics for severity of sleep disordered breathing.
2. Epidemiology and Risk Factors Overall OSA is a substantially more prevalent disorder than CSA with a ratio of between 10:1 and 20:1, although relative prevalence varies as a function of other co-morbid conditions. For example, CSA is relatively common in patients with Congestive Heart Failure. Perhaps the most widely quoted study on the prevalence of SDB is the Wisconsin Sleep Cohort Study which, in 1993, reported that 4% of middle-aged men and 2% of middle-aged women had an AHI of >5 events per hour, with associated daytime sleepiness. However, these estimates are dependent on the criteria employed. If the requirement for daytime sleepiness is removed, which is reasonable given that there is no evidence to suggest that the cardiovascular consequences of SDB are dependent on the presence of daytime sleepiness, the prevalence jumps to 24% and 9% for men and women respectively. Correspondingly, prevalence falls if the threshold AHI is increased and alternatively rises if disorders such as heavy snoring and upper airway resistance syndrome (subtle respiratory events not meeting criteria for hypopnea, also known as Respiratory Effort Related Arousals – RERAs) are included as components of SDB. Regardless, the prevalence of the disease is high and has likely risen since 1993, due to more sensitive techniques for measuring respiration, the aging of the population, and the ongoing obesity epidemic. The prevalence of SDB varies as a function of gender, being generally higher in males, although this finding may in part reflect clinical presentation and diagnostic factors. Ontogenetic development also has a major effect on SDB. Prevalence is relatively high in infants and young children, falls over infancy and adolescence, and progressively increases again as a function of age in adults. In women, the menopause represents a major risk factor for the occurrence of OSA. Prevalence in older adults (>65 years) remains uncertain with some studies suggesting an increase and others a decrease in the age-related effect on SDB. There are also complex interactions between gender and age. Thus, prevalence is higher in male than female infants, equivalent between the genders during childhood, higher in males through adolescence and adulthood, with the difference narrowing following menopause in women. These changes reflect changing etiological mechanisms. Indeed, it can be argued that SDB is not the same disorder in various age groups. The issue of mechanism will be considered further below. There is also preliminary evidence to suggest racial differences, with prevalence being higher in African-Americans and Asians, possibly originating from cranio-facial factors. Finally, it would be anticipated that the prevalence of SDB would change as a function of community characteristics and, for example, is likely to have increased as a function of the ongoing obesity pandemic. Thus, the estimated prevalence derived from the Wisconsin Sleep Cohort Study is likely to be close to a lower bound, with the upper bound estimate depending on the precise definition of SDB and the techniques used to assess its presence. In addition to gender, age and race, there are a number of additional risk factors for the disorder. The most influential is the relationship between OSA and obesity. Both population and cross sectional studies indicate a progressive increase in the prevalence of OSA with an increase in Body Mass Index (BMI). The distribution of fat deposition is important, with fat deposited in the abdomen, upper body and neck (android obesity) being more closely associated with OSA, a relationship that contributes to the high prevalence of OSA in middle aged men and indicates why neck circumference may be a better predictor of OSA than BMI. Craniofacial factors are also critical and while there are a large number of anatomical features that contribute, they all ultimately act to reduce the size or increase the collapsibility of the airway lumen. Broadly they divide into two classes: skeletal and upper airway soft tissue. The most frequently observed skeletal deviations are a short or retro posed mandible and inferiorly positioned hyoid bone. Quite subtle deviations, reflecting normal variations in the population, can predispose individuals to OSA, while substantial deviations associated with specific craniofacial disorders, such as Alpert’s Syndrome for example, produce very high prevalence. OSA is also associated with soft tissue structures, such as the tonsils, soft palate, uvula and tongue, that are enlarged relative to the size of the airway. These types of soft tissue abnormalities, especially tonsillar enlargement, are the most common underlying factor in children with OSA. The size of the airway lumen may also be reduced through lateral parapharyngeal wall thickness as a consequence of increased fat pad size secondary to obesity, or to a range of other mechanisms independent of obesity. There are a number of specific diseases that are associated with relatively high levels of SDB. Indeed, OSA is likely to be potentiated by any disorder that can narrow the airway lumen structurally, such as hypothyroidism, acromegaly, Down’s Syndrome and so forth, or functionally, such as through impaired upper airway muscle function following stroke. CSA also occurs at higher levels in specific disorders, such as in congestive heart failure.
3. Pathophysiology There is a general consensus that sleep apnea is due to an interaction between upper airway luminal size and instability in respiratory control. Luminal size may be considered to be the product of relatively stable anatomical factors, such as amount of soft tissue as a result of obesity or craniofacial disorders (see previous section), and functional properties, such as state related changes in upper airway muscle activity. The importance of different mechanisms varies over individuals, with respiratory control factors being predominant in CSA, and factors affecting airway size becoming more important in OSA. However, OSA and CSA pathogenesis have clear overlap as will be discussed. 3.1 Structural Considerations The importance of structural factors in the pathophysiology of OSA has been highlighted by the earlier section on epidemiology and risk factors. To reiterate, structure predisposes OSA by producing a narrow collapsible airway, as a consequence of the amount of soft tissue relative to the size of the craniofacial cavity. A narrow airway requires a large pressure gradient along the airway with marked down-stream negative pressure being produced by the diaphragm during inspiration. This negative pressure tends to reduce the size of the airway further and, if not opposed by upper airway dilator muscle activity, can result in flow limitation, but typically not closure of the airway. Further, a narrow lumen implies high extraluminal pressure (pressure derived from tissue surrounding the airway) at the site of collapse. Airway occlusion tends to occur at the end of exhalation or occasionally during early inspiration. The cause of collapse of the airway is embodied in the concept of the critical closing pressure, or Pcrit. The Pcrit concept has two components. First, the critical pressure determining whether the airway closes is the up-stream, not down-stream, pressure. Second, the airway will close when the up-stream pressure is less than the pressure difference across the lumen walls at the point of subsequent collapse. Thus, Pcrit is the up-stream pressure at which the airway collapses. The difference across the lumen walls is referred to as the transmural pressure and is defined as, “inside minus outside”, a greater negative value favoring collapse. Some authors also refer to a tissue pressure which predisposes the airway to collapse. Tissue pressure is an extra-luminal positive pressure and may be referenced to intra-luminal pressure or atmospheric pressure, more commonly the latter. Thus, one can consider an airway more vulnerable to collapse when transmural pressure is negative or tissue pressure is more positive. The internal up-stream pressure can be increased (made more positive) by Continuous Positive Airway Pressure (CPAP), thus reducing the likelihood of collapse, or decreased (made more negative) by negative pressure at the nares, increasing the likelihood of collapse. However, under normal physiological conditions it is the airway transmural pressure at the point of collapse that determines airway patency. The transmural pressure may be increased (made more negative) by abundant extra-luminal soft tissue (outside positive pressure relative to atmosphere), or by respiratory muscle activity reducing intra-luminal pressure (inside negative pressure). In contrast, propensity for airway collapse can be reduced by the activity of upper airway dilator muscles, or by stiffness in the luminal walls, where stiffness partly reflects structural properties and partly longitudinal tracheal traction. When the result of these forces is airway collapse (up-stream pressure falls below the extra-luminal positive pressure), the up-stream pressure at which this occurs is Pcrit. Airway obstruction rarely occurs during wakefulness, implying that forces opposing the transmural pressure are adequate, even in the presence of deficiencies in anatomy. This concept has been referred to as the “Balance of Forces”. It follows that, to prevent airway collapse, upper airway dilator muscle activity must be greater in individuals with anatomical deficiencies. However, as will be discussed in greater detail below, sleep is associated with a reduction in central drive to the respiratory system (loss of the wakefulness stimulus), including drive to the upper airway muscles. In OSA patients this state related fall in pharyngeal dilator muscle activity is sufficient to eliminate the capacity of the airway muscles to oppose transmural pressure resulting in pharyngeal collapse. It has also been assumed that in some individuals functional components may be more critical. For example, the magnitude of the fall in upper airway muscle activity during sleep may be excessive, or the behavior of the muscle impaired. However, evidence in support of the pathophysiology on this side of the equation has been slow to accumulate. 3.2 Respiratory Control Our understanding of respiratory control is based on four concepts: feedback loops, loop gain, the hypocapnic apnea threshold, and the wakefulness stimulus. The regulatory control of blood gases (CO2 and O2) depends on feedback from receptors. The location of these receptors, particularly the central receptors for CO2, results in feedback delays, with the inherent potential for instability, particularly when the delay is prolonged. Viewed another way, if the chemoreceptors were located in the lung, unstable ventilation would be improbable. The magnitude of the response to a perturbation in the system, its sensitivity, varies both within and between individuals. The term, loop gain, describes this relationship, formally being defined as the ratio of the magnitude of the response to a perturbation and the magnitude of the initial perturbation. When loop gain is >1, the response is greater than the perturbation and the system is unstable, resulting in alternating periods of hyper and hypoventilation. There are two major components to loop gain: controller gain, essentially the response of the ventilatory system to hypercapnic and hypoxic stimulation (i.e. chemoresponsivness); and plant gain, how effectively a given level of ventilation removes CO2, greater effectiveness being higher plant gain. The critical point being that, given the inherent delays in feedback control, the respiratory system becomes unstable, with fluctuations in ventilation, if loop gain exceeds one. Instability in respiratory control is primarily expressed as fluctuations in the activity of two effector systems, respiratory muscles, resulting in variations in ventilation, and upper airway dilator muscles, resulting in changes in airway luminal size and Pcrit. Depending on the weighting of these two components within an individual, respiratory instability may be expressed as an obstructive, mixed, or central event. The role of upper airway muscles in influencing Pcrit and thus the occurrence of obstructions has been discussed in the previous section. The effect of fluctuations in respiratory drive in central apnea is determined by the hypocapnic apnea threshold and the wakefulness stimulus. Ventilation tends to cease when CO2 levels fall below a threshold level (the so called “apnea threshold”). Under conditions of ventilatory instability the hyperventilation phase can drive CO2 below the apnea threshold, resulting in a central apnea. However, in cases where upper airway mechanics are particularly unfavourable, obstructive apnea could result, emphasizing the overlap in the pathogenesis of OSA and CSA. These apneas become repetitive because of the repetitive cycles of hyper-hypoventilation that are characteristic of unstable control. Instability in the control system is more marked in sleep. During wakefulness the apnea threshold is masked by a central tonic input to the respiratory system, referred to as the wakefulness stimulus. However, at sleep onset the withdrawal of the wakefulness stimulus unmasks both the apnea threshold and the underlying control instability. One implication of this is that sleep-wake state instability can itself be a source of respiratory instability. Several forms of CSA, CSA at altitude, Cheynes-Stokes ventilation in congestive heart failure patients and idiopathic central sleep apnea, have in common high loop gain as a consequence of high controller gain, which periodically drives CO2 below threshold. Other disorders, such as central alveolar hypoventilation, are characterized by waking hypercapnia.
4. Consequences In OSA subsequent morbidity is considered to be the result of three features of the disorder: the high negative intra-thoracic pressure that occurs as a consequence of respiratory effort against an occluded airway; intermittent hypoxia as a result of apneas; and because resolution of the obstruction requires an arousal from sleep in the majority of patients, repetitive arousal from sleep. These have a number of pathophysiological consequences. Two of these, increased sympathetic tone and endothelial dysfunction, result in an increased risk for a number of cardiac conditions. These include systemic and pulmonary hypertension, stroke, heart failure, cardiac arrhythmias, and coronary artery disease. Neuroendocrine function is disrupted resulting in metabolic dysfunction, particularly in glucose metabolism, resulting in insulin resistance and potential risk of worsening type 2 diabetes. There is also emerging evidence that, in addition to its role in causing OSA, obesity may be a consequence of sleep disruption, particularly sleep deprivation. Sleep fragmentation from repetitive arousals from sleep results in excessive daytime sleepiness, performance deficits and increased risk of traffic and industrial accidents. Finally, neuronal damage, probably secondary to intermittent hypoxia, leads to neurocognitive impairment. In many cases the relationship appears to be bi-directional, such as with obesity. The specific contribution of intra-thoracic pressure, intermittent hypoxia or arousal from sleep to specific disorders is not clearly understood. The pathophysiological consequences of CSA are less well understood, probably because of the lower prevalence of the disorder. The hemodynamic consequences appear less severe, although cardiac arrhythmias are present, as in OSA. CSA patients are more likely to complain of insomnia, rather than hypersomnia, perhaps as a consequence of a disruption to normal sleep patterns, and as with OSA, depression or fatigue is a common complaint. Congestive heart failure has a strong association with a particular form of CSA, Cheynes-Stokes respiration, which has a characteristic waxing and waning ventilation pattern. In this situation CSA is likely to be a consequence of the congestive heart failure, with some evidence to suggest that prognosis is poorer in heart failure patients with CSA. However, while treatment of the Cheynes-Stokes CSA improves cardiac function in congestive heart failure patients, it does not appear to improve life expectancy above that achieved by standard medical therapy including beta-blockers.
5. Diagnosis The standard diagnostic test for sleep apnea is overnight polysomnography in a clinical sleep laboratory. The usual measures taken include an electroencephalogram, electro-oculogram, electromyograms (chin and tibialis anterior), electrocardiogram, pulse oximetry, and measures of ventilation (thermistor or nasal pressure, for example) and snoring. As briefly discussed in the first section of this report, despite refinement over time, technical and definitional difficulties remain. However, space does not allow a presentation of the recommended measurement and definitions of the different types of SDB and the reader is referred to the recently released American Academy of Sleep Medicine manual (see Bibliography). Two central controversies will, however, be considered. A major issue is whether diagnostic recordings should be collected through laboratory based polysomnography or home monitoring. The major impetus for home monitoring comes from the cost and poor availability of in-laboratory polysomnography. However, home monitoring has an inherent problem. The accuracy and complexity of home monitoring systems are inversely related. Cost efficient systems are relatively inaccurate and are specific to OSA, while with the addition of further channels of information accuracy is improved, the complexity and cost approach in-laboratory polysomnography. In a similar vein, oximetry has been proposed as a screening device for OSA, but as with other simple forms of home monitoring, its sensitivity and specificity are highly variable across different studies. A further economic-resource issue is whether diagnosis and CPAP titration should occur on the same night; what are referred to as “split studies”. The first half of the night is used to make a diagnosis and where necessary the CPAP level appropriate for treatment is determined in the second half. This is compared to conducting the diagnosis on night 1 and titration on night 2. In general the literature does not suggest substantial disadvantages to split night studies, although minimal research has addressed this issue.
6. Management Continuous positive airway pressure, or CPAP, remains the most widely applied treatment. As noted under Pathophysiology, it is effective because it elevates up-stream pressure above the extra-luminal positive pressure at the potential point of collapse. If used by the patient, the efficacy of CPAP is very high with improvements in the pathophysiological consequences of the disorder, including sleepiness, neurocognitive impairments and, in patients with hypertension, BP. However, as implied, patient adherence is low. Adherence appears to be a trade-of between the patient’s perception of the benefits (primarily improvement in sleepiness and daytime functioning) and the deleterious side effects (a range of factors including dry mouth, mask soreness and psychosocial factors). Thus, more insidious consequences of OSA, such as cardiovascular and cognitive impairments that do not have obvious daytime effects, may not positively affect adherence. Attempts to improve adherence initially focused on the technology, including humidifying the air and various modifications of the CPAP flow generator. Recently there has been increasing focus on cognitive/behavioral approaches. Behavioral approaches also potentially provide an alternative to CPAP when the severity is mild. A reduction in obesity increases lumen size, sleeping prone or laterally, rather than supine, reduces the likelihood of the tongue falling back into the airway, and abstinence from depressant drugs such as alcohol improves upper airway muscle tone. However, weight reduction, avoiding sleeping supine and abstaining from alcohol themselves have low success rates and while behavioral approaches can be effective with some patients, they do not have wide applicability. The aim of surgery is to increase the size of the upper airway lumen either by removing soft tissue, the most usual approach, or by advancing the mandible to increase the size of the boney structure surrounding the airway and soft tissues. Several techniques have been developed to remove soft tissue including uvulopalatopharyngoplasty (UPPP), LASER and radio-frequency ablation techniques. When applied non-specifically to OSA, surgery has only limited efficacy. Similarly, surgery has been of only partial effectiveness in treating individuals for snoring, when they do not obstruct. To the extent that surgery has a role in the treatment of SDB it is likely to be in those patients who have specific and surgically correctable structural abnormalities. Moreover, we currently lack a reliable means to predict successful surgical responses. Oral devices vary in the sophistication of their technology but are, in general, designed to advance the mandible and prevent retroglossal collapse. The specific efficacy of oral devices is less than CPAP, particularly for severely affected patients. However, they have the distinct advantage that some patients prefer oral appliances to CPAP yielding improved treatment adherence. It is a possibility that the overall efficacy of oral devices may be equivalent to CPAP, particularly for mild to moderate disease. Oral appliances are a potential treatment method for those patients who have failed or refused CPAP. There is not an over-arching treatment for CSA and recommended treatments vary with the origin of the central apneas. In some patients treatment of the underlying disorder may be sufficient. For example, such an approach may be best for Cheynes-Stokes ventilation in congestive heart failure patients. Thus, this option should be the first to be considered. Several approaches have been used to treat CSA directly. CPAP has been effective for patients that have a reflex inhibition of ventilation in response to airway obstruction and may be effective for Cheynes-Stokes ventilation. In the former group of patients the CPAP prevents the initial collapse. In Cheynes-Stokes, CPAP may retard the development of hypocapnia, or may have a number of hemodynamic benefits, but the mechanism by which this is achieved remains uncertain. Administration of oxygen has been effective in some instances, although again the mechanism remains unclear. Pharmacological approaches have also been developed, primarily to alter CO2 control. Thus, for example, acetazolamide, which produces a metabolic acidosis and lowers the CO2 apnea threshold, is somewhat effective for various forms of central apnea including Cheyne-Stokes breathing. Finally, in patients with severe central alveolar hypoventilation or respiratory neuromuscular disorders with nocturnal hypoxia and hypercapnia, ventilatory support during sleep (using bi-level positive airway pressure) may be necessary.
7. Bibliography Caples, S.M., Garcia-Touchard, A. and Somers, V.K. (2007).Sleep disordered breathing and cardiovascular risk. Sleep, 30, 291-303. Eckert, D.J., Jordan, A.S., Merchia, P. and Malhotra, A. (2007). Central sleep apnea: pathophysiology and treatment. Chest, 131, 595-607. Iber, C., Ancoli-Israel, S., Chesson, A.L. and Quan, S.F. The AASM Manual for the Scoring of Sleep and Associated Events. American Academy of Sleep Medicine: Westchester,IL., 2007. Leung, R.S. and Bradley, T.D. (2001). Sleep apnea and cardiovascular disease. American Journal of Respiratory and Critical Care Medicine, 164, 2147-65. Pack, A.I. (2006). Advances in sleep disordered breathing. American Journal of Respiratory and Critical Care Medicine, 173, 7-15. Redline, S., Budhiraja, R., Kapur, V. et al., The scoring of respiratory events in sleep: reliability and validity. Journal of Clinical Sleep Medicine, 2007, 3, 169-200. White, D.P. Pathogenesis of obstructive and central sleep apnea. American Journal of Respiratory and Critical Care Medicine, 172, 1363-1370. Young, T., Palta, M., Dempsey, J., Skatrud, J., Weber, S. and Badr, S. (1993). The occurrence of sleep-disordered breathing among middle-aged adults. The New England Journal of Medicine, 32, 1230-1235. Young, T., Peppard, P. and Gottlieb, D. (2002). The epidemiology of obstructive sleep apnea: a population health perspective. American Journal of Respiratory and Critical Care Medicine, 165, 1217-1239.